Pith. sign in

REVIEW 6 major objections 5 minor 23 references

A frozen quantum Ising reservoir, watched through entropy and Fisher information, detects hidden regime shifts in non-stationary systems and gates a generative readout to suppress trajectory collapse.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-02 07:43 UTC pith:EI7RVIRY

load-bearing objection The central empirical claim is undercut by the paper's own Table 1: nHQRC is worse than the unmanaged drift baseline on every trajectory metric, and phase classification is essentially chance. the 6 major comments →

arxiv 2607.16281 v1 pith:EI7RVIRY submitted 2026-07-09 quant-ph cs.LGq-fin.CPq-fin.ST

A Novel Hybrid Quantum Reservoir Computing (nHQRC) for Phase Transition Detection in Non-Equilibrium Dynamical Systems

Manoj B. Bhatkar , Prashant M. Yawalkar This is my paper
classification quant-ph cs.LGq-fin.CPq-fin.ST
keywords hybrid quantum reservoir computingphase transition detectionvon Neumann entropyquantum Fisher informationStochastic Schrödinger Bridgetransverse-field Ising modelregime switchingnon-equilibrium dynamics
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes a hybrid quantum-classical pipeline for detecting latent phase transitions in highly non-stationary stochastic systems. Instead of training a variational quantum circuit, it feeds data into a frozen disordered Transverse-Field Ising Model reservoir, which acts as a nonlinear projection into an exponentially large Hilbert space. It tracks the von Neumann entropy and exact mixed-state Quantum Fisher Information of the reservoir as 'entanglement witnesses' that spike when the driving field is about to undergo a regime shift. Those spikes gate a generative Stochastic Schrödinger Bridge readout, suppressing drift and preventing trajectory collapse during chaotic episodes. In an 8-dimensional hidden-Markov regime-switching test, the framework improves drift-to-diffusion efficiency and arrests maximum trajectory decay by over 13% relative to a classical support-vector readout, with O(1) temporal overhead per step.

Core claim

The central claim is that a fixed, disordered Transverse-Field Ising Model quantum reservoir, combined with entropy and Quantum Fisher Information triggers and a Stochastic Schrödinger Bridge generative readout, can detect structural phase transitions in high-dimensional non-equilibrium systems and stabilize trajectories that classical regressors let collapse. The paper reports that in an 8-dimensional regime-switching benchmark, the nHQRC-SSB variant achieves 51.81% phase classification accuracy (vs 45.78% for classical SVR), a positive Matthews correlation of 0.0245 (vs -0.1068), improves drift-to-diffusion efficiency from -0.61 to -0.41, and reduces maximum trajectory decay from -72.35% t

What carries the argument

The load-bearing mechanism is the entanglement-witness gate: the von Neumann entropy S(ρ_A) of the reduced reservoir state and the exact mixed-state Quantum Fisher Information F_Q(ρ,J_z), computed with respect to the collective spin operator, serve as leading indicators of structural collapse. The entropy value S_t is fed into the Stochastic Schrödinger Bridge diffusion equation as an exponential damping factor exp(-2S_t) on the drift and a suppression factor (1-S_t) on the diffusion, so that when the reservoir detects a phase transition, the readout contracts the trajectory and holds it in a neutral preservation state. A pre-amplification manifold-scaling layer (θ = tanh-scaled and bounded

Load-bearing premise

The framework assumes that von Neumann entropy and Quantum Fisher Information of the driven reservoir, plus a threshold anchored to the 65th percentile of the full observed entropy series (Section VI-B), are causally available leading indicators of hidden regime switches in real time; the percentile threshold as described cannot be computed online without lookahead.

What would settle it

Run a strictly causal version of the pipeline where the entropy gate threshold is computed from a rolling window of past entropy values only, keeping all other settings fixed. If the maximum trajectory decay improvement over the classical SVR baseline falls below the reported 13%—or if the QFI's 1-interval leading cross-correlation of 0.76 drops to zero—then the claimed early-warning and shielding advantages depend on lookahead and disappear under real-time constraints.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the entropy and QFI triggers are genuine leading indicators, the framework offers a NISQ-compatible early-warning system for regime shifts in high-dimensional systems, requiring only O(1) temporal overhead per step.
  • The entropy-gated generative readout provides a 'safety floor' that can prevent catastrophic amplitude collapse during chaotic episodes, a behavior classical covariance-based models cannot reproduce.
  • The architecture sidesteps barren-plateau trainability issues by freezing the quantum reservoir and optimizing only classical readout and gate parameters.
  • The reported dimensionality crossover (classical methods remain superior for d≤4, quantum reservoir projection becomes necessary around d≥8) defines a practical domain of applicability for hybrid quantum reservoir computing.
  • The reported 1-interval leading cross-correlation of 0.76 between Quantum Fisher Information and volatility shocks suggests QFI can serve as a predictive, not just descriptive, entanglement witness.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The 65th-percentile execution gate is defined over the full observed entropy series in Section VI-B; a causal streaming implementation would need to estimate that percentile online, and whether the 13% improvement survives with a causal threshold is a direct test of the 'lookahead-free' claim.
  • If the entropy and QFI triggers are merely coincident transforms of the input, the framework's 'phase transition shield' reduces to a nonlinear gating rule; a randomized-shuffle test (shifting the trigger times) would separate causal early-warning from post-hoc correlation.
  • The O(1) temporal overhead claim covers only per-step evolution; the genetic optimization and full density-matrix tomography scale exponentially with qubit count, so a hardware deployment would need to show that entropy/QFI extraction remains practical beyond n=8.
  • The single benchmark uses a synthetic 8-dimensional regime-switching process with known ground truth; applying the pipeline to real-world non-stationary data with unknown transition times would test whether the QFI leading correlation persists outside synthetic settings.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

6 major / 5 minor

Summary. The paper proposes nHQRC, a hybrid quantum reservoir computing architecture that embeds stochastic driving fields into a disordered Transverse-Field Ising Model (TFIM) reservoir, extracts von Neumann entropy and mixed-state Quantum Fisher Information (QFI) as entanglement witnesses, and uses a classical Stochastic Schrödinger Bridge (SSB) readout with entropy-triggered gating. The authors claim that, on an 8-dimensional hidden Markov regime-switching benchmark, nHQRC-SSB improves drift-to-diffusion efficiency and arrests maximum trajectory decay by over 13% compared to standard classical benchmarks, and that QFI provides a 1-interval leading cross-correlation of 0.76. They further claim an O(1) temporal overhead suitable for NISQ hardware. The central evidence is Table 1, which compares Unmanaged Stochastic Drift, Classical SVR-SSB, and 8-qubit nHQRC-SSB on phase classification accuracy, MCC, RMSE, net state expansion, drift-to-diffusion efficiency (η), and Maximum Trajectory Decay (MTD). The paper includes a detailed Algorithm 1 specifying the pipeline.

Significance. If the claims were supported, the proposal would be significant: a NISQ-compatible reservoir that detects latent phase transitions before structural breakdown, with a generative readout that stabilizes trajectories, would address a real need in non-equilibrium dynamical systems. The architecture is specified in sufficient detail to permit independent implementation, and the use of entropy/QFI as dynamical triggers rather than passive feature extraction is a legitimate and interesting idea. However, the central empirical claim is contradicted by the paper's own Table 1: nHQRC-SSB has worse η and MTD than the Unmanaged Stochastic Drift baseline, while phase classification accuracy and MCC are statistically indistinguishable from chance. No code, data, confidence intervals, or repeated-experiment statistics are provided. The paper's contribution is therefore not established, and the internal contradiction goes to the core of the claimed 'phase transition shield.'

major comments (6)
  1. [Table 1; Section VI-B; Fig. 4 caption] The central performance claim is contradicted by the paper's own numbers. For the 8-qubit nHQRC-SSB row, η=-0.41 and MTD=-59.21%, both worse than the 'Unmanaged Stochastic Drift' baseline (η=-0.28, MTD=-57.14%). The improvement over Classical SVR-SSB (η=-0.61, MTD=-72.35%) is not an improvement over a standard classical benchmark or over doing nothing. The Fig. 4 caption's claim that the quantum-gated strategy 'significantly outperforms the unmanaged drift baseline' is therefore false on the reported metrics. Since the entropy gate is the mechanism claimed to 'actively arrest maximum trajectory decay,' this contradiction collapses the headline result.
  2. [Table 1] Phase Classification Accuracy = 51.81% and MCC = 0.0245 for the 8-qubit model. For a binary hidden-regime detection problem this is indistinguishable from chance. No confidence intervals, significance tests, or multiple-seed statistics are provided. Thus the 'phase transition detection' component of the central claim is unsupported, even setting aside the drift/diffusion metrics.
  3. [Section V-B] The genetic algorithm is 'designed to penalize Maximum Trajectory Decay (MTD)' and selects φ=0.7 using an in-sample window. Table 1 then reports MTD as an out-of-sample outcome. Because the same objective is used for selection, any MTD improvement is biased in favor of the selected configuration; reporting this metric as evidence of trajectory-preservation ability is circular. An independent held-out metric or a comparison across several non-selected hyperparameter settings is required.
  4. [Section VI-B] The execution gate is 'anchored to the 65th percentile of observed von Neumann entropy.' A percentile of the full observed series cannot be known at time t without future data, contradicting the 'lookahead-free' claim in the Abstract and Section IV-A. If the percentile is computed online from a running window or a cumulative distribution, the update rule must be specified. Without a causal threshold, the claimed early-warning 'phase transition shield' is not demonstrated; it may be a post hoc nonlinear transform of the same data.
  5. [Section IV-D vs Algorithm 1 (line 19)] The SSB update equation is inconsistent between the main text and the algorithm. Section IV-D gives dx_t = [μ_target exp(-2S_t) - x_t]/(1-τ) dt + sqrt(ε(1-S_t)) dW_t, while Algorithm 1 line 19 gives dx_t = μ_target exp(-2S_t)(1-x_{t-1})(1-τ) dt + dW_t · ε(1-S_t). These are different SDEs. Additionally, τ is used both as the TFIM evolution time in Algorithm 1 line 8 and as the bridge terminal time in the SSB equation. The readout is not reproducible without resolving this ambiguity.
  6. [Section IX; Section VI-B; Algorithm 1 steps 9-11] The paper claims an O(1) temporal overhead, but computing exact mixed-state QFI and von Neumann entropy requires the full density matrix eigensystem. On classical hardware this is exponential in n; on NISQ hardware full state tomography is not O(1). Section IX itself acknowledges an exponentially prohibitive classical bottleneck for n≥8, undermining the O(1) statement. The paper needs a concrete resource estimate for the tomography/eigensolver steps before this scalability claim can be accepted.
minor comments (5)
  1. [Section V-A / Fig. 1] The text contains the placeholder 'Error! Reference source not found.' after Fig. 1, indicating an unfinished cross-reference.
  2. [Table 1] The metrics η and MTD are not defined mathematically anywhere in the paper. Without definitions or units, the table cannot be independently reproduced.
  3. [General] No error bars, confidence intervals, or repeated-run statistics are reported. All evaluations appear to be single realizations of a stochastic process, so no statistical significance can be assessed.
  4. [General] No code or data repository is provided, despite the synthetic benchmark being fully specified. A public implementation would be needed for the O(1)-overhead and phase-detection claims to be verifiable.
  5. [References] Some references appear to be future-dated or otherwise unverifiable (e.g., [10], [21]), and citation formatting is inconsistent. Please check all entries against current database records.

Circularity Check

2 steps flagged

nHQRC's headline MTD improvement is the GA fitness objective, and its entropy execution gate is anchored to the full observed entropy series, so the central 'phase-shield' predictions reduce to in-sample constructions.

specific steps
  1. fitted input called prediction [Section V-B (Genetic Meta-Optimization) and Table 1 (Out-of-Sample Metrics)]
    "Governed by a trajectory efficiency fitness function designed to penalize Maximum Trajectory Decay (MTD), the evolutionary loop autonomously localized and locked the global coupling strength at φ = 0.7. ... Table 1 lists 'OUT-OF-SAMPLE METRICS UNDER 8-D REGIME-SWITCHING DYNAMICS' and reports 'Maximum Trajectory Decay (MTD) -57.14% -72.35% -59.21%'."

    The Hamiltonian coupling φ (and the GA-discovered architecture W_GA) are selected by minimizing MTD on an in-sample window, and then the very same MTD family is reported as nHQRC-SSB's headline out-of-sample gain ('recovering over 13% of the systemic amplitude decay'). The paper documents no held-out split separating the fitness data from the evaluation data, so the reported trajectory-preservation advantage is the optimized objective rather than an independent prediction.

  2. self definitional [Section VI-B ('The QFI Coincident Witness Print and Phase Transition Shield') and Algorithm 1, Phase 3]
    "By anchoring the active execution gate to the 65th percentile of observed von Neumann entropy, the framework significantly improved the systemic drift-to-diffusion efficiency (η) from -0.61 (Classical SVR benchmark) to -0.41 (nHQRC-SSB)."

    The 65th-percentile threshold is defined from the observed von Neumann entropy series, and the same entropy series S_t is then used inside the SSB readout as the dynamic execution gate (dx_t contains exp(-2S_t) and (1-S_t)). Thus the gate is defined in terms of the very trajectory statistics used to evaluate the 'phase transition shield'; the claimed leading/early-warning behavior is a post-hoc construction from the full in-sample series rather than a causal, lookahead-free prediction.

full rationale

The paper's strongest claim—that nHQRC-SSB 'actively arrests maximum trajectory decay'—reduces in part to the GA's own fitness function: the same MTD metric used to select φ/W_GA is reported as the out-of-sample improvement, with no documented held-out split. The entropy execution gate is likewise anchored to a percentile of the observed entropy series, making the 'leading quantum trigger' an in-sample statistic rather than an independent early warning. These two reductions are the basis for the score of 6. Separately, the paper contains a serious, non-circular internal contradiction: Table 1 shows nHQRC-SSB (η=-0.41, MTD=-59.21%) performing worse than the unmanaged drift baseline (η=-0.28, MTD=-57.14%), and phase classification (51.81%, MCC 0.0245) is near chance; this is a correctness/evidence problem, not a circularity, and it further weakens the headline claims. The authors' self-citations [16,17] are literature reviews and are not load-bearing for the nHQRC mechanism, so self-citation does not contribute to the score. No machine-checked proofs, code reproductions, or external benchmarks are provided to make the MTD/gate results independent of the fit.

Axiom & Free-Parameter Ledger

6 free parameters · 6 axioms · 0 invented entities

The central method relies on several fitted hyperparameters (φ, ζ, entropy-gate percentile, SSB constants, SVR settings, data-generator parameters) and on domain assumptions about quantum witnesses and O(1) tomography that are not independently evidenced. No new physical entities are introduced.

free parameters (6)
  • TFIM coupling matrix / φ = φ=0.7; W_GA=[0.293, 1.352, 1.122, 0.847, 1.247, 1.330, 1.197, 0.200]
    Genetic algorithm tunes the coupling matrix on an in-sample simulation window using a trajectory-efficiency fitness function that penalizes MTD; the reported trajectory improvement is measured on the same objective.
  • Amplification factor ζ = 12.0
    Empirically locked by the optimization loop; ζ=20.0 produced entropy plateaus. It directly controls rotation angles and is fitted to the same benchmark.
  • Entropy gate percentile = 65th percentile of observed von Neumann entropy
    Threshold for the active execution gate is anchored to the observed entropy series and used to classify phase transitions on the same series.
  • SSB drift/diffusion parameters τ, ε = not reported
    Appear in the SSB SDE and control trajectory contraction and noise; values are not specified or fitted with a stated procedure.
  • SVR readout hyperparameters = not reported
    The classical drift projection uses a Support Vector Regressor; kernel, regularization, and window length are unspecified, making the classical baseline irreproducible.
  • Synthetic benchmark generator parameters = not reported
    Hidden Markov generator matrix, drift/volatility matrices, jump amplitude J, and Poisson intensity define the benchmark; without them no independent replication of Table 1 is possible.
axioms (6)
  • domain assumption The reservoir state under the frozen TFIM acts as a natural kernel with inner product K(u,u')=|⟨ψ(u)|ψ(u')⟩|²
    Section III-A; no proof that the TFIM feature map is a useful or universal kernel for arbitrary stochastic inputs, and no comparison to standard QRC kernels is provided.
  • standard math Pezzè-Smerzi criterion F_Q(ρ,J_z)>N certifies multipartite entanglement
    Section III-B; this is a standard result used to interpret QFI spikes, but the interpretation as a leading phase-transition indicator is the paper's own.
  • ad hoc to paper Spikes in von Neumann entropy and QFI are leading indicators of macroscopic phase transitions
    Core trigger assumption in Section VI-B; supported only by a coincident cross-correlation of 0.88 and a 1-interval correlation of 0.76 on the same synthetic series, not by an independent causal or out-of-sample test.
  • ad hoc to paper Full density-matrix tomography and exact eigen-decomposition for QFI are feasible at O(1) temporal cost on near-term hardware
    Sections IV-C and IX: the paper admits classical simulation is exponentially prohibitive and asserts native hardware execution solves it, but state tomography and eigensystem extraction on a physical register are not O(1).
  • ad hoc to paper The SDE in Section IV-D is a valid Stochastic Schrödinger Bridge producing correct trajectory distributions
    No derivation or citation to Schrödinger bridge theory is given; the equation in Section IV-D differs from the Appendix version and lacks existence/uniqueness assumptions.
  • domain assumption The 8-dimensional jump-diffusion HMM benchmark is representative of non-equilibrium dynamical systems
    Section VI; synthetic data with chosen generator matrix, heavy-tailed jumps, and cross-correlations is used to make general claims about fluid, financial, and quantum systems.

pith-pipeline@v1.3.0-alltime-deepseek · 11726 in / 16932 out tokens · 159477 ms · 2026-08-02T07:43:54.650676+00:00 · methodology

0 comments
read the original abstract

The analysis of highly non-linear stochastic data within non-equilibrium dynamical systems requires computational frameworks capable of detecting latent phase transitions before systemic structural breakdowns occur. Traditional Variational Quantum Algorithms (VQAs) are frequently bottlenecked by vanishing gradients, the barren plateau problem, and prohibitive training overheads. In this paper, we propose a novel Hybrid Quantum Reservoir Computing (nHQRC) framework, which bypasses these limitations by employing a frozen, disordered Transverse-Field Ising Model (TFIM) to project time-dependent stochastic driving forces into an exponentially large Hilbert space. To resolve the physical phase multi-wrapping vulnerabilities present in baseline quantum reservoir models, we introduce a lookahead-free, pre-amplification manifold scaling technique. Multi-qubit configurations are genetically optimized to the "edge of chaos," while quantum state tracking is performed by extracting von Neumann entropy ($S$) and exact mixed-state Quantum Fisher Information (QFI) to act as leading entanglement witnesses. Utilizing these quantum triggers as boundary constraints, trajectory predictions are constructed via a generative Stochastic Schr\"odinger Bridge (SSB) readout. By subjecting the quantum reservoir to an 8-dimensional non-stationary stochastic driving field, the framework significantly improves systemic drift-to-diffusion efficiency ($\eta$) and actively arrests maximum trajectory decay (MTD) by over 13% compared to standard classical benchmarks. This establishes a robust, $\mathcal{O}(1)$ temporal overhead blueprint for near-term quantum regime detection and macroscopic subsystem stabilization.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

23 extracted references · 1 canonical work pages

  1. [1]

    Quantum computational finance: Monte Carlo pricing of financial derivatives,

    P. Rebentrost, B. Gupt, and T. R. Bromley, “Quantum computational finance: Monte Carlo pricing of financial derivatives,” Phys. Rev. A , vol. 98, no. 2, p. 022321, 2018, doi = 10.1103/PhysRevA.98.022321. https://link.aps.org/doi/10.1103/PhysRevA.98.022321

    doi:10.1103/physreva.98.022321 2018

  2. [2]

    Solving the optimal trading trajectory problem using a quantum annealer,

    G. Rosenberg, P. Haghnegahdar, P. Condrat, and A. Golselmes, “Solving the optimal trading trajectory problem using a quantum annealer,” IEEE J. Sel. Top. Signal Process., vol. 10, no. 6, pp. 1053– 1060, Sep. 2016, doi: 10.1109/JSTSP.2016.2574703. https://arxiv.org/pdf/1508.06182

    arXiv 2016

  3. [3]

    Quantum Reservoir Computing Using Jaynes -Cummings Model

    S. Das, Giorgi, G L, and Zambrini R. "Quantum Reservoir Computing Using Jaynes -Cummings Model." ArXiv, (2025). https://arxiv.org/abs/2510.00171

    Pith/arXiv arXiv 2025

  4. [4]

    Quantum machine learning in feature Hilbert spaces,

    M. Schuld and N. Killoran, “Quantum machine learning in feature Hilbert spaces,” Phys. Rev. Lett., vol. 122, no. 4, p. 040504, 2019, doi: 10.1103/PhysRevLett.122.040504. https://link.aps.org/doi/10.1103/PhysRevLett.122.040504

    doi:10.1103/physrevlett.122.040504 2019

  5. [5]

    Quantum Reservoir Computing: A Reservoir Approach toward Quantum Machine Learning on Near-term Quantum Devices,

    K. Fujii and K. Nakajima, “ Quantum Reservoir Computing: A Reservoir Approach toward Quantum Machine Learning on Near-term Quantum Devices,” ArXiv, (2020). https://arxiv.org/abs/2011.04890

    Pith/arXiv arXiv 2020

  6. [6]

    Dynamic portfolio optimization with real datasets using quantum processors and quantum -inspired tensor networks,

    S. Mugel, K.Carlos, E. Sánchez et al., “Dynamic portfolio optimization with real datasets using quantum processors and quantum -inspired tensor networks,” Phys. Rev. Appl., vol. 4, no. 1, p. 013006, 2022, doi: https://doi.org/10.1103/PhysRevResearch.4.013006 https://link.aps.org/doi/10.1103/PhysRevResearch.4.013006

    doi:10.1103/physrevresearch.4.013006 2022

  7. [7]

    Quantum machine learning,

    Biamonte, J., Wittek, P. Pancotti, N. et al . “Quantum machine learning,” Nature, vol. 549, pp. 195 –202, Sep. 2017, doi: 10.1038/nature23474. [Online]. Available: https://www.nature.com/articles/nature23474

    doi:10.1038/nature23474 2017

  8. [8]

    PennyLane: Automatic differentiation of quantum circuits,

    V. Bergholm et al., “PennyLane: Automatic differentiation of quantum circuits,” arXiv preprint arXiv:1811.04968 , 2018. https://arxiv.org/abs/1811.04968

    Pith/arXiv arXiv 2018

  9. [9]

    Scikit -learn: Machine learning in Python,

    F. Pedregosa et al. , “Scikit -learn: Machine learning in Python,” J. Mach. Learn. Res. , vol. 12, pp. 2825 –2830, Nov. 2011. https://www.jmlr.org/papers/volume12/pedregosa11a/pedregosa11a.p df

    2011

  10. [10]

    A Quantum Reservoir Computing Approach to Quantum Stock Price Forecasting in Quantum -Invested Markets

    Otieno, W., Zagoskin, A., Balanov, A. G., Gongora, J. T., and E., S, “A Quantum Reservoir Computing Approach to Quantum Stock Price Forecasting in Quantum -Invested Markets” ArXiv. Feb. 20 26, https://doi.org/10.48550/arXiv.2602.13094 https://arxiv.org/abs/2602.13094

    Pith/arXiv arXiv doi:10.48550/arxiv.2602.13094

  11. [11]

    Optimizing a Quantum Reservoir Computer for Time Series Prediction

    Kutvonen A, Sagawa T, and Fujii K. "Optimizing a Quantum Reservoir Computer for Time Series Prediction." ArXiv, (2018). https://arxiv.org/abs/1807.03947

    Pith/arXiv arXiv 2018

  12. [12]

    Quantum reservoir computing for realized volatility forecasting ,

    Q. Li, C. Mukhopadhyay, A. Bayat and A.Habibnia , “Quantum reservoir computing for realized volatility forecasting ,”Phys. Rev. Research, vol 8, no.3, p. 023028, Apr. 2026, https://link.aps.org/doi/10.1103/rbj7-4wnq

    doi:10.1103/rbj7-4wnq 2026

  13. [13]

    Theory of overparametrization in quantum neural networks,

    M. Larocca, N. Ju, D. García -Martín, P. J. Coles, and M. Cerezo, “Theory of overparametrization in quantum neural networks,” Nature Comput. Sci., vol. 3, no. 6 , pp. 542–551, 2023, doi: 10.1038/s43588 - 023-00467-6. https://www.nature.com/articles/s43588-023-00467-6

    doi:10.1038/s43588 2023

  14. [14]

    Bhatia, Matrix Analysis, New York, NY, USA: Springer -Verlag, 1997

    R. Bhatia, Matrix Analysis, New York, NY, USA: Springer -Verlag, 1997

    1997

  15. [15]

    Adiabatic quantum computing,

    T. Albash and D. A. Lidar, “Adiabatic quantum computing,” Rev. Mod. Phys., vol. 90, no. 1, p. 015002, Jan. 2018, doi: 10.1103/RevModPhys.90.015002 https://link.aps.org/doi/10.1103/RevModPhys.90.015002

    doi:10.1103/revmodphys.90.015002 2018

  16. [16]

    A review of hybrid quantum -classical methods,

    M. Bhatkar and P. Yawalkar, “A review of hybrid quantum -classical methods,” JETIR, 2025. https://www.jetir.org/papers/JETIRHA06027.pdf

    2025

  17. [17]

    Predictive techniques for stock price movement using quantum -classical approach: A comprehensive review,

    M. Bhatkar, P. Yawalkar and V. More, “Predictive techniques for stock price movement using quantum -classical approach: A comprehensive review,” IJSET, 2025. DOI: https://doi.org/10.5281/zenodo.17961944 https://zenodo.org/records/17961944

    arXiv 2025

  18. [18]

    Hybrid quantum -classical reservoir computing of thermal convection flow,

    P. Pfeffer, F. Heyder, and J. Schumacher, “Hybrid quantum -classical reservoir computing of thermal convection flow,” Physical Review Research, vol. 4, no. 3, p. 033176, Sep. 2022. https://link.aps.org/doi/10.1103/PhysRevResearch.4.033176

    work page doi:10.1103/physrevresearch.4.033176 2022

  19. [19]

    Hybrid quantum -classical reservoir computing for simulating chaotic systems,

    F. Wudarski, D. O'Connor, S. Geaney, A. A. Asanjan, M. Wilson, E. Strbac, P. A. Lott, and D. Venturelli, “Hybrid quantum -classical reservoir computing for simulating chaotic systems,” arXiv preprint arXiv:2311.14105v2, Apr. 2024. https://arxiv.org/abs/2311.14105

    Pith/arXiv arXiv 2024

  20. [20]

    Memory -Augmented Hybrid Quantum Reservoir Computing,

    J. Settino et al., “Memory -Augmented Hybrid Quantum Reservoir Computing,” arXiv preprint arXiv:2409.09886v2 , Nov. 2024. https://arxiv.org/abs/2409.09886

    Pith/arXiv arXiv 2024

  21. [21]

    Hybrid Photonic Quantum Reservoir Computing for High-Dimensional Financial Surface Prediction,

    F. Amanov and A. Azamov, “Hybrid Photonic Quantum Reservoir Computing for High-Dimensional Financial Surface Prediction,” arXiv preprint arXiv:2603.10707v1 , Mar. 2026. https://arxiv.org/html/2603.10707v1

    arXiv 2026

  22. [22]

    Quantum Encoding and Analysis on Continuous Time Stochastic Process wi th Financial Applications,

    X.-N. Zhuang, Z. -Y. Chen, C. Xue, Y. -C. Wu, and G. -P. Guo, “Quantum Encoding and Analysis on Continuous Time Stochastic Process wi th Financial Applications,” arXiv preprint arXiv:2208.02364v5, Sep. 2023. https://doi.org/10.22331/q-2023-10- 03-1127

    Pith/arXiv arXiv 2023

  23. [23]

    Quantum reservoir computing in atomic lattices,

    Llodrà G., Mujal P., R. Zambrini, G. L. Giorgi, “Quantum reservoir computing in atomic lattices,”Chaos, Solitons & Fractals, Volume 195,2025 , https://doi.org/10.1016/j.chaos.2025.116289

    arXiv 2025